Artificial intelligence tools are part of how we work and part of what we build for clients. This page sets out our position on AI clearly: how we use it in our own work, what we do and do not use it for, how we handle data when AI tools are involved, and how we think about the environmental cost of AI adoption.

We are publishing this because we think clients deserve to know. Transparency about AI use is increasingly important and we would rather set out our position plainly than leave it unaddressed.

How We Use AI in Our Own Work

We use AI tools as part of our development and delivery process in a considered way. They are tools, not replacements for professional judgement. Where AI assists our work, a qualified engineer or specialist reviews, validates and takes responsibility for the output. Nothing produced with AI assistance goes to a client without that review.

Where AI helps us

  • Code assistance. AI tools help with boilerplate, pattern completion and exploring approaches. All generated code is reviewed, tested and understood before it is used in production.
  • Research and summarisation. AI assists with processing information quickly. We verify sources and do not rely on AI-generated summaries for anything where accuracy is critical.
  • Documentation. AI assists with first drafts of technical documentation. All documentation is reviewed and edited before publication.
  • Testing. AI tools assist in generating test cases and identifying edge cases we might otherwise miss.

Where we do not use AI

  • We do not use AI to generate client deliverables without disclosure. If AI has played a meaningful role in producing something for a client, we say so.
  • We do not use AI to make architectural or security decisions. These require professional judgement and accountability that AI tools cannot provide.
  • We do not input client data, client code or sensitive business information into public AI tools. Where AI tooling requires data to be processed externally, we use only tools with appropriate data handling agreements in place.
  • We do not present AI-generated content as human-produced work where the distinction matters.

Data Privacy and AI

This is the area where AI use carries the most risk for businesses, and where we are most cautious.

Public AI tools, including widely used chat and code assistant products, process inputs on remote servers. Data entered into these tools may be used to train future models, retained by the provider, or accessible to third parties depending on the product’s terms. Entering personal data, confidential client information, proprietary business logic or sensitive commercial data into public AI tools is a data governance risk that most organisations have not fully assessed.

Our approach is to treat public AI tools as we would any other third-party service: with a clear view of what data is appropriate to share and what is not. We do not enter client data into public AI tools. For client projects that involve AI integration, we select and configure tools with data privacy as a primary constraint, scoped from the start.

Where clients ask us to advise on their own AI use, data privacy is one of the first things we address. Our AI awareness training for business teams covers this directly.

AI and Environmental Impact

The energy consumption of AI is a real and underreported issue. Training large language models requires enormous computational resources and generates significant carbon emissions. Running AI tools at scale, even through APIs, consumes meaningful amounts of energy that is rarely visible to the end user.

We are conscious of this in how we approach AI adoption. We use AI tools where they improve outcomes, and we are deliberate about not reaching for them by default when a simpler approach will do the same job. More processing is not always better, and we do not treat AI as a default layer to add to every problem.

When advising clients on AI integration, we raise the environmental dimension alongside the practical and commercial ones. That might mean recommending a smaller, more targeted model over a larger general-purpose one, discussing the energy implications of running AI inference at the proposed scale, or flagging that a proposed AI feature will add infrastructure cost and carbon cost without a proportionate benefit.

This position is set out in more detail on our sustainability page, which covers how FullyCoded approaches environmental responsibility across the business.

AI in Client Projects

When we build AI into a client product or platform, we follow the same principles that govern our own use: appropriate data handling, professional oversight of AI outputs, transparency with the client about what is being used and how, and a clear view of whether AI is the right tool for the job.

We do not recommend AI integration because it is fashionable. We recommend it where it solves a real problem, handles a task that would otherwise require significant manual effort, or creates a capability the product genuinely needs. Where it does not, we will say so.

For clients who want to understand AI better before committing to integration, our AI for Business consultancy and our AI awareness training are both designed to give businesses the grounding they need to make informed decisions rather than following industry momentum.

Keeping This Policy Current

AI tools and the landscape around them are changing quickly. We will review and update this policy as our practice evolves, as new tools become relevant, and as the regulatory and data governance environment develops. The date below indicates when this version was last reviewed.

Last reviewed: May 2026